Core MLflow Components
- MLflow Tracking: Log parameters, metrics, code versions, and artifacts from Python training scripts using
mlflow.log_params()andmlflow.log_metrics(). - MLflow Models: Package models into standard flavors (Scikit-learn, PyTorch, ONNX) usable across serving APIs and batch jobs.
- Model Registry: Centralized model store managing versions, stage transitions (Staging ──► Production), annotations, and approval gates.